18
might give state or provincial governments a greater say in the allocation of budgets to capital
expenditures, even when that authority is not reflected in financial terms. To the extent that
state governments can internalize the fiscal costs of expenditure, we expect this dummy
variable to have a negative effect. However, if fiscal management in a federation is
characterized by bailouts and moral hazard problems between states and the federation (as in
Brazil until quite recently), the variable could have a positive effect on pork. The data is
taken from Henisz’s “Political Constraints” dataset.
F) Lower chamber seats. The count of legislative seats in the lower house (lowseat) is
a discrete variable, which seeks to contrast our results with those in Crain and Bradbury
(2001). Data is taken from (and cross-referenced between) Political Handbook of the World
(various issues) and PoliSci.com.
H) Upper chamber seats. The count of legislative seats in the upper house (upseat) is
also a discrete variable. This variable refines the Bicameralism variable above, and similarly
tests Crain and Bradbury’s argument that upper house seats reduce particularistic spending.
Data is taken from PoliSci.com.
G) Number of districts. We include a variable on the number of districts in the lower
chamber (ldistr) to test whether the common pool problem is created by the number of
legislators (as argued by Crain and Bradbury, 2001) or the number of distinct electoral
districts. We believe that the number of districts better captures the common pooling
problem under the law of 1/n, and expect the variable to have a positive effect on
particularistic spending. The data was obtained from the “Database of Political Institutions
3” and Political Handbook of the World.
Convention